Transforming Research, Risk Assessment and Compliance

156

By Wael Jadallah, General Manager, APAC and Middle East, and Dimitrios Papanastasiou, Head of Industry Practice Group and Head of Digital & Advanced Analytics, Moody’s

Generative artificial intelligence (GenAI) is poised to revolutionize the banking industry, offering unprecedented opportunities for efficiency, innovation and value creation. As financial institutions race to adopt and scale this transformative technology, GenAI is reshaping core functions, such as research, risk assessment and compliance.

By leveraging large language models (LLMs) and traditional artificial intelligence (AI)/machine learning (ML) capabilities, banks can unlock new insights, streamline complex processes and enhance decision-making across their organizations. This article explores how GenAI is transforming key banking functions and examines specific use cases that demonstrate its game-changing potential.

AI can unlock opportunities for banks across multiple workflows. In particular, GenAI could transform a commercial bank’s lending/credit-lifecycle process, including assessing customers’ creditworthiness using various metrics, rating/scoring and credit approval, alongside know-your-customer (KYC) and anti-money-laundering (AML) credit checks.

Transforming financial research and analysis

One of the most promising applications of GenAI in banking is to enhance research and analysis capabilities. The technology can dramatically improve the speed, depth and scale of financial research, empowering analysts to generate insights and make decisions with unprecedented efficiency.

For example, Moody’s has developed a GenAI-powered research assistant that leverages extensive proprietary data and content to help users rapidly synthesize information and uncover insights. It can analyze multiple documents about a company, sector or country and provide concise summaries in seconds—a task that would take human analysts hours or even days to complete manually. And in terms of efficiency gains? The time it takes to write a comprehensive credit summary can, in some cases, be reduced by more than 60 percent.

Some key capabilities of GenAI-enhanced research tools include:

• Rapidly extracting and synthesizing key information from lengthy financial reports, regulatory filings and other documents,
• Generating comprehensive company profiles that combine financial data, credit ratings, sustainability metrics and qualitative insights,
• Providing industry and macroeconomic analyses by consolidating information from multiple expert sources,
• Answering complex queries by drawing connections across vast datasets.

For example, an analyst could ask the system to summarize a company’s financial performance and its strengths and weaknesses compared to its peers, credit-rating drivers, sustainability profile and prospects based on recent corporate actions and industry outlooks. A GenAI assistant can compile this holistic view in seconds, allowing the analyst to focus on higher-level analysis and decision-making.

Importantly, leading GenAI research tools can go beyond simply retrieving information to generate novel insights. By integrating GenAI and ML-based predictive analytics, these systems can run scenario analyses and forecasts. An analyst could ask how a company’s credit rating and key financial ratios would change if interest rates rose by 100 basis points, for instance.

Enhancing risk assessments and early-warning systems

Risk assessment and monitoring are critical functions for banks, and GenAI promises to enhance these capabilities significantly. By analyzing vast amounts of structured and unstructured data, GenAI-powered systems can identify potential risks earlier and with greater accuracy than traditional methods.

Financial institutions are developing next-generation early-warning solutions that leverage GenAI to transform the credit-monitoring process. These tools can consolidate disparate data sources—including financial statements, behavioral data, macroeconomic indicators, news sentiment and more—to provide a holistic view of risk.

Some key applications in risk assessment include:

• Automated financial analysis and commentary generation: GenAI can analyze a company’s financial statements and produce well-articulated summaries of performance, allowing risk analysts to focus on evaluating creditworthiness and refining recommendations.

• Enhanced early-warning signals: By training on historical performance data and incorporating multiple risk factors, GenAI models can uncover subtle patterns that indicate increased risk. The models can also stress test an obligor’s repayment capacity and provide forward-looking risk assessments.

• Portfolio-level insights: GenAI assistants allow risk managers to analyze entire portfolios quickly, identifying concentration risks, emerging trends and potential problem areas through natural language queries (NLQs).

• Remediation recommendations: When risks are identified, GenAI systems can suggest proactive measures, such as adjusting credit limits or restructuring debt. They can also generate customized communications to at-risk borrowers.

For example, a credit-risk manager could ask the GenAI system to identify the riskiest segments of the loan portfolio, taking into account recent macroeconomic developments. Incorporating the latest economic data and forecasts, the system would analyze the entire portfolio, highlight vulnerable areas and explain the key risk drivers.

Transforming KYC and AML processes

Know-your-customer and anti-money-laundering processes are crucial for banks but often involve time-consuming manual work. GenAI can significantly enhance these functions by automating information gathering, risk profiling and investigative processes.

While AI and machine learning techniques are already used in areas such as transaction monitoring and fraud detection, GenAI opens up new possibilities, particularly for deeper investigations and risk profiling.

Some key applications include:

• Intelligent KYC assistants: GenAI can power conversational interfaces that guide compliance officers through the KYC process, helping them verify entity information, update risk profiles and conduct screening checks.

• Enhanced entity resolution and risk profiling: By analyzing data from multiple sources—including company registries, adverse media, sanctions lists and ownership databases—GenAI can create comprehensive risk profiles for individuals and entities.

• Automated investigation support: For AML investigations, GenAI can collate relevant information, suggest lines of inquiry and even draft investigative reports.

• Policy and regulatory updates: GenAI can help compliance teams stay up to date on evolving regulations by summarizing key changes and their implications for KYC/AML processes.

Financial intelligence (FININT) providers are exploring how GenAI can enhance KYC and compliance solutions, including company databases and risk-intelligence platforms. By combining vast amounts of structured data with GenAI’s natural language capabilities, these tools aim to provide compliance officers with more intuitive and powerful investigative capabilities.

Implementation challenges and best practices

While the potential of GenAI in banking is immense, successful implementation requires careful planning and governance.

Some key considerations include:

• Operating model: Industry research suggests that banks with centralized GenAI operating models are seeing the best results so far. A higher percentage of banks with highly centralized GenAI approaches have progressed to putting use cases into production compared to those with decentralized models.

• Risk management: Banks need robust frameworks to manage GenAI-specific risks, such as hallucinations (false or illogical outputs), data-privacy issues and potential biases. This includes implementing appropriate controls, monitoring systems and establishing governance processes.

• Data strategy: High-quality, well-organized data is crucial for effective GenAI implementations. Banks need to invest in data infrastructure and governance to leverage the technology fully.

• Talent and skills: Building GenAI capabilities requires a mix of technical expertise and domain knowledge. Banks need strategies to attract AI talent, upskill existing employees and foster collaboration between technical and business teams.

• Ethical considerations: As GenAI systems take on more decision-making roles, banks must ensure they are used ethically and transparently, with appropriate human oversight.

What’s ahead?

The potential is enormous, and so, too, are the ways in which to improve existing systems with GenAI to develop new solutions. Nevertheless, we are still talking about a technology that only recently gained mainstream acceptance.

Will the efficiency gains in consolidating information across different sources be overshadowed by the efforts required to fine-tune prompt engineering or develop a large language model from scratch? Will automated generation replace human intervention, or will the output not meet quality standards and require extensive human oversight? The results so far are encouraging, but for a highly regulated industry like banking, further evidence is needed to ensure robust, accurate and unbiased input into the decision-making process.

In the meantime, institutions need to invest in enterprise-wide education around AI and GenAI. They must also rely on verifiable data and information to power the generative capabilities of AI instead of open-source alternatives, design new operating models to allow for a controlled mix of human and technological interactions and put in place the appropriate level of governance oversight for GenAI models’ outputs.

One of the biggest challenges is to ensure the ethical and responsible use of this technology and safeguard the integrity of its output by imposing appropriate guardrails. This area is still rapidly evolving, and the output has not been consistently tested and used in decision-making.

We foresee institutions setting up governance teams as a matter of priority, especially in such a heavily regulated sector as banking. The industry as a whole is making quick strides in developing comprehensive validation frameworks for the outputs of GenAI models.

As more foundational GenAI models (pre-trained models) are developed, we will see a wider variety of AI options, each trying to find the perfect mix of size, clarity, adaptability and efficiency. Such progress will provide tangible and significant business benefits, transforming the banking industry.